Season naming and the local environment

Seasonal patterns vary dramatically around the world, and we explore the extent to which systems of season categories support efficient communication about the local environment. Our analyses build on a domain-general information-theoretic model of categorization across languages, and we identify several qualitative predictions that emerge when this model is applied to season naming, including the prediction that systems with even numbers of categories should be more common than systems with odd sizes. We test the model quantitatively using a collection of season systems drawn from the linguistic and anthropological literature and data specifying temperature and precipitation in locations associated with these systems. Our results support the predicted even-odd asymmetry, and we also find that the model makes a number of successful predictions about the locations of boundaries between seasons.